Anirn. Behav., 1985, 33, 620-627

The timing of decisions during shell investigation by the hermit crab, Pagurus bernhardus

ROBERT W. ELWOOD & A N N E S TEWA RT

Department of Zoology, The Queen's University of Belfast, Beljast, Northern Ireland

Abstract. Factors which influence decisions by hermit crabs concerning whether to approach, investigate and enter another shell have been investigated by systematically varying the size of the shell in possession and the size of the shell being offered. The probability that a crab will approach or enter the shell depends on both variables. Investigation after contact, however, depends only on the size of the shell in possession. Durations of each stage of the sequence are negatively correlated with the numbers of crabs in each experimental group performing the next stage. For the final sequence of aperture investigatory activities, the fewer crabs entering the offered shell, the greater the number of investigatory acts performed by those crabs. In addition to variation in number of acts and duration of investigation, specific use of appendages varies according to the experimental situation. Thus naked crabs use the minor cheliped to investigate shells in preference to the major cheliped, which is normally used by housed crabs, with the major cheliped being reserved for defence by naked crabs. These data are discussed in terms of information collection and decisions made on the basis of that information.

Animals select their habitat according to a variety of criteria. These criteria may be relatively simple, as in the settling of marine invertebrate larvae, or may be more complex, such as territory selection by birds (see Partridge 1978 for a review). In matters such as where to build a nest, a variety of sites may be inspected and a decision made, presumably on the basis of certain cues which should indicate that selection of one site is more likely to result in higher fitness than selection of another. Site selection is inferred from the non-random distribution of the animals but the cues that are involved are fre- quently not obvious (Partridge 1978).

Hermit crabs show preferences for particular types of habitat (Reese 1969) but within the habitat they have to make further decisions concerning which gastropod shell to occupy. Hermit crabs of several species show distinct preferences for certain species of shell (Reese 1963) and within a particular species of shell, specific sizes relative to crab size are preferred (Reese 1962). Such shell preferences have also been shown in Pagurus bernhardus, the subject of the present study.

In general, shell species or size preferences may be demonstrated either by the willingness of ani- mals to (a) orient towards other shells (Reese 1963), (b) enter an empty shell (Reese 1963; Elwood et al. 1979), (c) engage in a shell fight with another crab (Hazlett 1970; Dowds & Elwood 1983) or (d) escalate a shell fight (Dowds & Elwood 1983). In

each case the crab presumably assesses the quality of the alternative shell compared with that in possession. This occurs for the attacking crab in shell fights (Elwood & Glass 1981) and it has been demonstrated that the duration of the shell assessment process during a shell fight is influ- enced both by the quality of the shell in possession and by the difficulty of discriminating which is the better shell (Dowds & Elwood 1983). During a shell fight, however, other features are also being assessed, e.g. size and reactions of the defending crab (Dowds & Elwood 1985), and these interact with and influence the shell assessment process.

Thus, in order to examine the shell assessment process in the present study we have observed solitary crabs investigating empty shells. Crabs may be induced to occupy shells which, although of the same species, differ in size and thus suitability. The suitability of the shell in possession may reasonably be expected to influence appetitive behaviour for new shells and the tendency to change shells if a new one is located. When an empty shell is located it may be assessed by the crab and on the basis of information concerning both the shell in possession and that being examined the crab makes a decision on whether or not to move into the new shell. This study examines these two sources of information and their roles in the timing of decisions in the shell selection process.

620

Elwood & Stewart." Hermit crab decisions 621

M E T H O D S

The crabs, all Pagurus bernhardus occupying Lit- torina littoralis shells, were collected f rom a single beach on the N o r t h Down coast of N o r t h e r n Ireland. In the labora tory the crabs were main- ta ined in large tanks con ta in ing well-aerated sea water. The tempera ture was kept at approximate ly 6~ and the hours of i l luminat ion were appropr ia te for the t ime of year. The crabs were fed on pieces of fish every few days.

The major i ty (91%) of L. littoralis shells (not count ing those conta in ing the mollusc itself) at the collection site conta in crabs (Elwood et al. 1979) and some crabs are forced to occupy sub-opt imal shells. Therefore a prel iminary experiment was

conducted in which individual crabs were given a choice of nine shells. The mean weights and volumes of the shells occupied over a 4-day period were calculated and used to determine regression lines between crab weight and bo th shell weight and internal shell volume. These regression lines were regarded as indicat ing the preferred weight and volume for each weight of crab and were used to determine the shells to be used in the main experiments (Dowds & Elwood 1983). In sub- sequent experiments, however, shell weight only was used to quantify shells as shell weight and volume are highly correlated (Elwood et al. 1979).

The crabs for the main experiment were removed from their shells by cracking the shells in a bench vice. Female crabs were not used so as to eliminate

Table 1. Behaviours recorded during observations on crabs offered a new shell

Behaviour Description

Precontact activities

Withdraw

Walk Stationary

Random

Contact

Antennal contact Leg contact

Cheliped contact

External exploration

Hold

Aperture exploration

Leg

Minor + leg

Major + leg

Both chelipeds + leg

Major Minor Both Shell entry

Crab deeply withdrawn into shell or naked crab curled up Ambulation (includes approach) Crab not walking or actively moving appendages, not withdrawn Crab not walking but actively moving appendages

Antenna(e) in contact with new shell Leg(s) in contact with new shell but not grasping shell Cheliped(s) in contact with new shell but not grasping shell Crab holding shell with walking legs, actively moving chelipeds over shell surface Crab grasps shell but no active investigation

One or more walking legs inserted into aperture One or more walking legs plus minor cheliped inserted into aperture One or more walking legs plus major cheliped inserted into aperture One or more walking legs plus both chelipeds inserted into aperture Major cheliped inserted into aperture Minor cheliped inserted into aperture Both chelipeds inserted into aperture Crab rapidly moves into new shell

622 Animal Behaviour, 33, 2

any influence of eggs being carried on the abdomen. The males were gently dried with a paper towel, weighed and then placed in a small crystallizing dish filled with sea water at 6~ Ninety crabs remained naked (0% group), 90 were given shells that were 50% of the preferred weight (as deter- mined in the preliminary experiment; 50% group), 90 were given shells that were 100% of the preferred weight (100% group) and 90 were given shells that were 150% of the preferred weight (150% group). All crabs offered shells entered those shells within 2 h, and most did so within a few minutes. The following day (16 18 h later) the crystallizing dish was placed behind a one-way Perspex partition and an empty shell (filled with sea water to expel the air) was placed aperture upwards in the middle of the crystallizing dish. The new shell was either 50%, 100% or 150% of the preferred weight for that crab. In this manner 12 experimental groups (in a 3 x 4 design) were established, each containing 30 crabs. Each crab was observed once and then returned to the shore. The weight range of the crabs was 0.0%0-56 g and the experimental groups were balanced as regards this variable. Data were col- lected for the groups concurrently to avoid any seasonal effects.

The activities of each crab were recorded over a 30-rain period or, if the crab was investigating the new shell, until the crab either entered it or rejected it. The data were recorded directly on to an ITT 2020 (Apple) computer and stored on floppy disk. The mutually exclusive activities recorded were as shown in Table I.

RESULTS AND D I S C U S S I O N

Of the 360 crabs observed, 292 (8 l%) approached and made contact with the offered shell, 258 (72%) initiated an active investigation of the shell and 211 (59%) entered the shell. The main decisions that crabs appear to be making are thus whether to (a) approach and make contact, (b) actively investi- gate and (c) enter the offered shell. In this section we examine the factors that appear to influence these three decisions, the main factors examined being the two major variables of the experiment: (i) the size of shell in possession and (ii) the size of shell on offer. All statistical tests were performed on raw data but percentages are given as the sample size varies from case to case (see above).

(a) P< 0'001 100-

50-

0% 50% t00% 150%

size of she[[ occupied

(b) P< 0"01

50% 100% 150%

size of shell offered

Figure 1. The percentage of crabs making contact with the offered shell depending on (a) the size of the occupied shell and (b) the size of the offered shell. Shell sizes are expressed relative to the size preferred by the crab.

Whether to Approach and Make Contact

There was a significant influence of the shell in possession (P<0.001; Z2=21.47, df=3) with naked crabs being the most likely to approach (94%) with fewer crabs per group approaching as the size of the shell in possession increased (Fig. la). If approach to another shell is a reflection of motivation to seek and change shells this is not the expected pattern. It would be predicted that crabs in their optimum shell size (100% group) would have the lowest motivation to change shells and thus the lowest motivation to approach other shells. That crabs in the 150% group approached the least (although not significantly less than the 100% group) suggests that the crabs in these excessively large shells may have a slightly reduced mobility due either to poor fit or excess weight.

The quality of the shell on offer also influences whether or not crabs approach (P<0.01; Z2= 10.26, df=2) with crabs being most likely to approach optimal shells and least likely to approach the 50% shells (Fig. lb). This indicates that the crabs must initiate their evaluation of shell quality prior to contact with the shell; presumably this initial, distant evaluation is visual. It appears to be similar to the initial evaluation of crab size observed in aggressive encounters (Dowds & Elwood 1983) and supports previous observations that crabs may visually detect the size of shells (Reese 1963).

Whether to Investigate

Crabs in optimal shells were the least likely to actively investigate the shell once contact had been

Elwood & Stewart. Hermit crab decisions 623

(a) P < 0.00I (b) NS 100

5 0

0 % 50% 100% 150% 5 0 % 100% 150%

slze of shelf occupied size of sheK offered

Figure 2. The percentage of crabs continuing to active investigation after making contact depending on (a) the size of the occupied shell and (b) the size of the offered shell. Shell sizes as in Fig. 1.

P <0-001 P<0-001 (a) (b)

IO0-

Z o

50- g

0

0 % 50% 100% 150% 50% 100% 150%

size of she[[ occupied size of she[[ offered

Figure 3. The percentage of crabs continuing to shell entry after active investigation depending on (a) the size of the occupied shell and (b) the size of the offered shell. Shell sizes as in Fig. 1.

made (P < 0.001; ,%2= 16.40, d r=3 ; Fig. 2a). This is in agreement with the suggestion that these crabs should have the lowest tendency to change, shells. However, there is no significant difference in active investigation dependent upon the quality of the shell on offer (Fig. 2b), presumably as no further assessment of shell quality has taken place after the initial visual assessment.

Whether to Enter the New Shell

Investigating crabs in optimal shells were the least likely to enter the offered shell whereas all naked crabs that investigated the offered shell entered it (P < 0.001; g 2 = 37.69, dr= 3; Fig. 3a); one naked crab entered the shell without active investi- gation. Crabs offered optimal shells (Fig. 3b) were the most likely to enter the offered shell whereas those offered 50% shells were the least likely ( P < 0.001; Z2 = 28.31, dr=2). The outcome of these

three main decisions may be seen in Fig. 4 which shows the number of crabs in each individual group which entered the offered shell. These data gener- ally confirm the trends seen from combined data in Fig. 3 based on shell entries by those crabs that investigated.

Timing of Decisions

Having shown that the shell in possession may influence whether crabs make contact with the shells, actively investigate them and enter them and that the shell on offer influences two of these decisions we will now examine the time taken to make these decisions. The numbers of crabs (out of 30) in each group that approached and made contact with the shell ranged from 17 to 29. These numbers presumably reflect the average motivat ion of the crabs to change shells based upon the quality of the shell in possession and the quality of the

30-

o 20. o

V_ o 10-

(5 c

0 size of shetL occupied 0% 50~176176

size of she[[ offered 50~

0% 50~176176 0% 50~ 100~176

100~ 150%

Figure 4. The number of crabs per group that entered the offered shell. Shell sizes as in Fig. 1.

624 Animal Behaviour, 33, 2

30- r= =0"68 P<O'02

20- �9 �9 ~d 2 �9

c ~ 10-

0 700 1400

median du ra t i on precontact ac t i v i t i es (s)

Figure 5. The number of crabs of each group that made contact with the offered shell plotted against the median duration of precontact activities.

alternative shell (as "perceived' at that time). If this is the case then the latency to make contact with the offered shell might be expected to reflect also the motivational state of the crabs at that time. The correlation between latency and number making contact is significantly negative ( r = - 0 . 6 8 P<0 .02 ; Fig. 5) indicating that the fewer crabs making contact the longer it took those that did to do so.

The initial contact with the shell is most likely to be with the distal port ion of the antennae and to a lesser extent with the tips of the chelipeds or walking legs. This normally precedes the active investigation during which the crab climbs on to the shell and grasps it whilst examining it with a series of movements of the chelipeds and legs. The percentage of those crabs that made contact and then continued to active investigation presumably reflects the motivation of the crabs at that time and

g

g

c

I00

50.

r=-0.77 P<:O'O05

10

median dura t ion init ial contact (s)

20

�9 Figure 6. The percentage of crabs continuing to active investigation plotted against the median duration of the initial contact.

100

50

r=-0.69 P<O,02

50

median duration active shell investigation (s)

100

Figure 7. The percentage of crabs that entered the offered shell plotted against the median duration of active investigation.

thus the duration of the initial contact activities might be expected also to reflect this motivational state. This presumption appears to be correct as there is a significant negative correlation between the percentage o f those that made contact and continued to active investigation and the duration of initial contact ( r = - 0 " 7 7 ; P < 0 " 0 0 5 ; Fig. 6).

Similarly the duration of active investigation might be expected to be negatively correlated with the percentage of animals that investigated and then entered the offered shell, and this is the case (r = - 0'67; P < 0.02; Fig. 7). This period of active investigation is often a complex sequence of activi- ties and is presumably the time when most informa- tion is gained concerning the comparison of the qualities of the shell in possession and that on offer. This sequence will thus be examined more closely.

Active investigation may be split into two dis- tinct subgroups: firstly, external examination and secondly, aperture investigation. Once again the durations of these activities might be expected to

100

SO

@

r= -0 .69 P<O.02

@

40 80

median durat ion aper tu re inves t iga t ion (s)

Figure 8. The percentage of crabs that entered the offered shell plotted against the median duration of aperture investigation.

Elwood & Stewart." Hermit crab decisions. 625

r = - 0 . 7 9 P < 0 . 0 0 5 100 -

5 0 -

e c �9 �9 o

0 7 14

mean no ac ts in a p e r t u r e expLorat ion

Figure 9. The percentage of crabs that entered the offered shell plotted against the mean number of aperture investigations.

reflect the motivational state of the animal as indicated by the percentage entering the new shell. For external examination, however, this is not the case and no significant correlation exists between this activity and the percentage entering the shell. This suggests that external examination provides information concerning discrete categories (such as whether it is a shell or not) rather than information concerning quality. The duration of aperture inves- tigation relative to the percentage entering the new shells, however, shows a significant negative corre- lation (r = - 0.69; P < 0-02; Fig. 8). Furthermore, if the number of acts used in the sequence of aperture investigation is examined in relation to the percent- age entering the new shells, a significant negative correlation is again found ( r= -0"79 ; P<0.005; Fig. 9). It thus appears that this sequence of activities provides the final information concerning the quality of the shell on offer compared with the one in possession.

Table II. The percentages of crabs using specific investigatory activities when offered shells of differ- ent sizes

Activity

Size of offered shell

50~, 1007o 150~o Pvalue

Majorcheliped+legs 0~ 9.9~ 18"5~o <0.01 Both chelipeds +legs 15~ 43~ 43~o < 0-01 Major cheliped 61~o 69~o 80~o < 0.05

P values relate to z2-tests performed on raw data.

We have shown that the timing of various decisions (e.g. latency to make contact, duration of initial contact and duration of aperture explo- ration) reflects the motivation of the crabs as assessed by the percentage going on to the next stage of the sequence. A close examination of Fig. 8, however, in which median duration of aperture investigation is plotted against the percentage of crabs that change, shows a number of inconsistent results. For example, in each of three groups 77~ of the crabs entered the offered shell yet the median duration of aperture investigation varied greatly (3.2 s, 14 s and 40 s for the 0~o 150~, 50~o-150~ and 50~o 100~ groups respectively where e.g. 0~o-150~o represents naked crabs offered a 150~o shell). The following section examines possible reasons for these and other inconsistencies in the data.

Use of Specific Appendages

Exploration of the aperture included several activities involving insertion of different thoracic

Table III. The percentage of crabs, housed in different sized shells, using specific investigatory activities when occupy- ing shells of different quality

Activity

Size of occupied shell

0~o 50~, 100~ 150~o P value

Minor cheliped+leg 0~o 14~o 21~o 44~o <0-00t Both chelipeds + leg 0~ 41~ 76~ 70~o <0.001 Major cheliped 33~ 95~ 90~ 95~ <0.01 Minor cheliped 47~o 66~o 48~ 70~,,, < 0.05 Both chelipeds 18~ 95~o 100~o 95~o <0-01

P values refer to Z2-tests performed on raw data.

626 Animal Behaviour, 33, 2

100 P<O-O01

&

S 50. 2 0 -4q m*lj. r~ln bath maj. rain both rnctj rain both moj rain. both

s i z e of o c c u p i e d 0 % 5 0 % 100% 150% she[[

Figure 10. The percentage of those crabs that in their final exploratory act used the major cheliped (maj.), the minor cheliped (min.) or both chelipeds together (both) is shown for each experimental group. Only data for crabs that entered the shell are given.

appendages into the shell aperture. The data were examined for differences between experimental groups in the use of different appendages by those crabs that moved into the new shell.

Significant differences occur in the use of the major cheliped, either alone or in combination with the legs or with the minor cheliped and legs (Table II): use increases as the size of the shell being investigated increases. If the use of the appendages is an attempt to assess the size of the aperture, then whether or not the major cheliped can be inserted may be a good guide as to the suitability of the shell. Thus these differences in appendage use, dependent upon the size of the offered shell, are to be expected.

It was not expected, however, to find differences in the utilization of appendages depending upon the quality of the shell in possession (Table III). In this case, however, most of the differences may be attributed to the very short investigations per- formed by naked crabs, thus resulting in low numbers of naked crabs using each combination. If only the final act preceding shell entry is examined this should overcome the problem of having some short sequences. In this case (Fig. 10) we still see marked differences between the groups. Crabs in shells tend to use the major cheliped, either alone or in combination with other appendages, in the last act before changing shells and rarely use the minor cheliped. Conversely, naked crabs use their minor chelipeds more often than they use their major chelipeds for their last act before changing. The reason for this may be due to the greater danger to naked crabs from potential attack. Thus they keep

their major cheliped ready for defence and assess the shell with the minor cheliped. This apparent timidity of naked crabs was also noticeable in the group that was offered shells larger than preferred. Only 23 out of 30 crabs entered these shells compared with 29 out of 30 for both groups of crabs offered either 50% or 100% shells. This suggests that naked crabs are wary of approaching and investigating shells that are large as they may contain a large crab. Large housed crabs have been shown to attack naked crabs (Dowds 1983). Thus the 0%-150% group appears to have fewer crabs entering the new shell (23/30) than would be expected by examination of the causal factors for changing shells. Apparently another motivational factor associated with fear is interfering with the process of shell investigation.

The finding that both the shell in possession and the shell on offer appear to influence the use of particular aperture investigatory acts may also produce apparent inconsistencies in the data, as different acts presumably differ in their informa- tion gathering utility and therefore the speed of decision. In general, however, there is reasonable agreement between the motivational state at any one time in the sequence and the speed with which the next stage is attained.

C O N C L U S I O N S

The assessment of the shell in possession involves a variety of sensory pathways (Reese 1963) and the resultant input should determine the command state (McFarland & Sibly 1975) for shell change. The term 'command state' is normally used to describe an internal (physiological) state but it is reasonable to use it to describe any deviation from an optimal state. The command state presumably remains stable during shell investigation but the assessed quality of the alternative shell or 'cue state' (McFarland & Sibly 1975) will change as informa- tion is gathered. If the joint output of cue and command space is high (above a certain threshold) we expect to see the crab move into the new shell. If the joint output is low (below a second threshold) we expect to see investigation cease. If the joint output is between these thresholds then some means of resolving the problem must occur to prevent continued shell investigation.

It is anticipated that the potential costs of changing shells will influence this decision-making

Elwood & Stewart." Hermit crab decisions 627

process (McCleery 1978). If these costs are low then the crab may benefit by changing shells if the output o f the command and cue spaces are between the thresholds, as crabs should gain a more accu- rate assessment of a shell by moving into it. I f that shell is now perceived as inadequate, the crab may return to the old shell. I f the potential costs are high, for example if predation risks are high or if other hermit crabs are present that may steal the original shell, then it presumably pays to cease shell investigation without moving into the shell. The thresholds may thus be adjusted according to cost assessment. That potential costs influence shell assessment behaviour during shell fights has recently been demonstrated (Dowds 1983). Berried females may lose eggs during a shell exchange (the eggs becoming dislodged from the abdominal appendages), and thus face higher costs. Such females spend longer than other animals in assess- ing opponent 's shells both prior to eviction of the opponent and after eviction but prior to changing shells. After changing shells they spend less time investigating the old shell than do other animals, presumably because of the greater degree of cer- tainty obtained by the initial longer investigation. That potential costs are taken into account has further been demonstrated in the present study in that naked crabs do not use the normally favoured major cheliped in shell investigation but hold this appendage in readiness for defence and use the minor cheliped for investigation.

The precise mechanism of decision-making dur- ing shell investigation has yet to be determined but is the subject of continued research. An approach currently employed is to offer shells with partially blocked apertures, making them totally unsuitable for habitation. Thus all crabs give up the investiga- tion. By varying the quality of the shell in posses- sion and a variety of external stimuli (other crabs or predators) the interaction of shell assessment and cost assessment may be examined. It is hoped that this approach will provide further information concerning both this decision-making process and

the mechanisms involved in habitat selection in general.

A C K N O W L E D G M E N T S

We thank Ian Montgomery, Sue Neil, Ian Sneddon and two anonymous referees for their comments on earlier versions of this paper.

R E F E R E N C E S

Dowds, B. M. 1983. Aggression in the hermit crab, Pagurus bernhardus. Ph.D. thesis, The Queen's Univer- sity of Belfast.

Dowds, B. M. & Elwood, R. W. 1983. Shell wars: assessment strategies and the timing of decisions in hermit crab shell fights. Behaviour, 85, 1-24.

Dowds B. M. & Elwood, R. W. 1985. Shell wars II: the influence of relative size on decisions made during hermit crab shell fights. Anim. Behav., 33, 64%656.

Elwood, R. W., McClean, A. & Webb, L. 1979. The development of shell preferences by the hermit crab Pagurus bernhardus. Anim. Behav., 27, 940-946.

Elwood, R. W. & Glass, C. W. 1981. Negotiation or aggression during the shell fights of the hermit crab Pagurus bernhardus? Anita. Behav., 29, 1239-1244.

Hazlett, B. A. 1970. The effect of shell size and weight on the agonistic behaviour of a hermit crab. Z. Tierpsy- chol., 27, 369-374.

McCleery, R. H. 1978. Optimal behaviour sequences and decision making. In: Behavioural Ecology: An Evolu- tionary Approach (Ed. by J. R. Krebs & N. B. Davies), pp. 377-410. Oxford: Blackwell.

McFarland, D. J. & Sibly, R. M. 1975. The behavioural final common path. Phil. Trans. R. Soc., 270, 265-293.

Partridge, L. 1978. Habitat selection. In: Behavioural Ecology: An Evolutionary Approach (Ed. by J. R. Krebs & N. B. Davies), pp. 351-376. Oxford: Blackwell.

Reese, E. S. 1962. Shell selection behaviour of hermit crabs. Anim. Behav., 10, 347-360.

Reese, E. S. 1963. The behavioural mechanisms underly- ing shelI selection by hermit crabs. Behaviour, 21, 78-126.

Reese, E. S. 1969. Behavioural adaptations of intertidal hermit crabs. Am. Zool., 9, 343-355.

(Received 13 January 1984; revised 19 March 1984; MS. number." 2473)